Tools of the trade: statistical analysis in dermatopathology articles

Statistical analysis of research results provides powerful tools for understanding the data from projects. A prospective review of the statistical methods utilized in 100 consecutive articles in the recent literature relevant to dermatopathology was performed. The majority of papers, 75%, did not contain statistical analyses. A wide variety of methods were utilized in the papers that did have statistical analyses including methods in the following categories: t-test, contingency tables, multiple regression, multiple comparisons, nonparametric tests, life tables, and survival tests. Nine of the 25 papers utilizing statistical analysis had problems in the methods. These problems included treating categorical data as a continuous variable, multiple comparisons, subgroup analysis, and discordance of statistics and conclusions. It is important to understand the underlying assumptions of statistical methods to use the appropriate tets. These are tools of the trade for dermatopathology investigators.